Re: Summary of GNOME Mobile GUADEC BOF

[Matthew Allum <mallum openedhand com>]

Hi;

Hi;

I unfortunately didn't make GUADEC this year, therefor nor the mobile
meeting :(. Many thanks for such an extensive and interesting summary.
I've added some quick thoughts from myself Id of hopefully raised if
there in person.

On Fri, 2008-07-18 at 14:39 -0700, Dave Neary wrote:
> 
> 
> The third part of the meeting was a fairly wide-ranging discussion on
> GNOME's place in the industry, particularly in light of recent "bad
> press" (or, at least, potential bad press). Chris Blizzard suggested
> that if Nokia were to change its licencing policy for QT that GNOME
> would then have some hard choices to make - a business friendly QT
> licence could help them supercede GTK+ as the toolkit of choice (Mark
> Shuttleworth supported this & added to it that he felt it would be
> entirely possible to have a GNOME experience built on QT if the
> licencing were right).

Yeah, though my thoughts are even if Nokia were to change the license to
something more liberal/business friendly GTK+ still has the advantage in
that its 'neutral' and not owned/controlled by a single entity. With
devices converging more and more and thus new competitors being formed
all the time, its questionable as to depend a major piece of technology
ultimately controlled by a competitor - especially if there are good
compelling alternates. 

My more extreme thoughts would be that for mobile at least it doesn't
really matter - Qt and GTK+ are dead there. The iphone came and made
them into antiques. Its now all about running the full UI on OpenGL to
be relevant. Neither of which GTK nor Qt can do nor were designed to do.
We have a lot of nice stuff below the UI in GNOME Mobile and I really
believe we're ahead by quite a margin in being able to build OpenGL
based UI's with GNOME Mobile technologies (obviously Im biased however).

> 
> There seemed to be some agreement that individually, OLPC and OpenMoko
> were both cases of projects which had larger problems than the toolkit -
> ever-changing direction, difficult competitive environment, ill-defined
> problem set - which were working against them. A few people (myself
> included, and Mark Shuttleworth) said that while it was tempting to
> discount these stories, that it would be a missed opportunity of taking
> a long hard look at the GNOME platform, and seeing what problems people
> are having with it, and work to fix it.
> 

Yes this sucks unfortunately and even more in the fact that *we* know
its not really a problem with the toolkit but seemingly to the outside
this may not be perceived as the case. I guess it really only becomes a
problem when either succeeds greatly with the alternate. Time will tell.

> Unfortunately (from my point of view) we didn't get a huge amount of
> traction on that discussion. Something to take up with the GNOME ISV
> community at large, perhaps? The basis for a roadmap for the platform?
> 
> (Aside: From a personal point of view, laying things out with Glade is a
> nightmare - GtkHBoxes inside VBoxes inside HBoxes... a nice flexible
> canvas with some HIG aware development tools for placing widgets like
> they have on the Mac would be wonderful. I know GObject used to hurt my
> brain too back in the day. But I haven't done enough development with
> the GNOME platform to comment on the ease of use of things like
> Telepathy, GStreamer, D-Bus, GConf, GTK+, ...)
> 
> One thing that was said is that the bindings (Python, Mono & Java were
> mentioned explicitly) suffer from leaky abstractions, and have both
> performance & quality issues. And a lack of binding-specific
> documentation. Currently you have to guess the names of methods &
> classes from the C namespace, more or less.

Vala, Vala, Vala! This really has a lot of potential on mobile if
someone really starts pushing it and proving it. It hopefully gives many
of the pluses of a scripting language but avoids the negatives you
mention above which become extra critical on low end platform.

> 
> On a positive note: it was reiterated that the GNOME platform has been
> extremely successful in being adopted - it's part of maemo, Ubuntu
> Mobile, LiMo, moblin, ALP, and is already being used and deployed in
> devices by a wide range of companies including Vernier, iRex, Garmin,
> Nokia, OpenMoko, OLPC (who haven't abandoned Sugar), and others I'm
> forgetting. We have an opportunity to communicate about GNOME Mobile
> being the common platform for a whole range of devices and initiatives,
> to show the world (press & application developers) that mobile linux
> isn't as fragmented as it might appear. And we have some new
> technologies like Clutter arriving which have tremendous potential to
> provide a rich 2.5D and 3D application framework to developers.

Yes yes !! :-)

I think the biggest downside to Clutter currently is there is nothing
(public) in terms of an application that really proves substantially
what it can do. Things are beginning to crop up though - Entertainer (a
media box type app) for example is looking really nice. Mobile h/w with
GL drivers etc is not all that accessible to o/s devs, but even thats
changing with things like the Zoom kit. Difficult underlying
infrastructure problems to be solved to of which the iPhone pretty much
cleverly works around (but I think that is beginning to show cracks now
3rd party apps). 

Another downside the fact that people keep using this oblique 2.5D term
in describing Clutter ;P That makes me think its some kind of Populous
style isometric game engine. It makes me cringe like being referred to
as 'Matt'. Do I really going to have to write some kind of crack 1980's
necromancer style virtual office type UI with Clutter to kill this
description ? ;-)

Anyway, I degress, what I mean to say is Im wanting to encourage people
to grab Clutter, play with it, hopefully like it and then build amazing
useful beautiful things with it. We're heavily focused at OH in
improving the Clutter core continually and will do our best to help
other building cool stuff with Clutter.

> 
> So - there lies the idea - gather a small group of people, focussed on a
> specific problem that is fixable (or, at least, which can be worked to a
> proof-of-concept design & code) after a 2 day hackfest. Ideas, suggestions?
> 

How about looking at hacking on something we're missing rather than
something that pretty much does the job (i.e gstreamer) - Things that
come to mind are syncing and filling holes around gsm telephony based
stuff ?

Hope that wasn't all a bit too random - its pretty late here.

Many thanks;

  == Matthew